Abstract:
The success of Turing Test technologies for system validation depends on the quality of the human expertise behind the system. As an additional source of human experts' validation knowledge a Validation Knowledge base (VKB) and so called Validation Expert Software Agents (VESAs) revealed to be useful. Both concepts aim at using collective (VKB) and individual (VESA) experience gained in former validation sessions. However, a drawback of these concepts were their disability to provide a reply to cases, which have never been considered before. The paper proposes a case-based data mining approach to cluster the entries of VKB and VESA and derive a reply to unknown cases by considering a number of most similar known cases and coming to a "weighted majority" decision. The approach has been derived from the k Nearest-Neighbor (k-NN) approach.